> for example, those who earn more tend to express more fear and anger on Twitter.
That is to be expected. If you know your r/K selection theory, K individuals have larger-than-average amygdalae which is responsible for awareness, risk-assessment and consequentially fear.
K individuals in human beings are those who aren't promiscuous, marry early and have kids, and by consequence are infinitely less likely to be poor.
> Perceived optimists have a lower mean income.
This also makes sense, since r-individuals in human beings are more welcome to hope and change (optimism), and they tend to be less well-off financially since they switch their survival needs onto K-individuals (demanding help from them for paying their student loans, providing food stamps, etc).
Something as benign as your user agent, previous browsing activity, or individual site cookie can predict income more accurately than a regressional analysis of salary data as compared to twitter themes.
